3.5 Fracture surface observation
4.1.2 Fatigue crack growth tests
The results of the fatigue tests (Figure 4.4 a) show that, under the same loading condition, panels made of AA2198 generally have higher fatigue resistance than the panels of the material AA2139. The fatigue life improvement by crenellations in the material AA2198 is two times of that in the material AA2139. The crack propagation rates calculated using a 5-points-polynominal fitting method [78] were plotted against the corresponding stress intensity factor ranges from the FEM
Figure 4.1. Microstructures of (a) AA2139 and (b) AA2198.
Figure 4.2. Variation of texture in different materials and at different positions of the panel.
4.1. Results 45
Figure 4.3. Color coded mapping of [100] inverse pole figures in both materials.
simulations in double logarithm form (Figure 4.4 b). The data points of both flat and crenellated panels of AA2139 fall into the same linear band as characterized by Paris Law whereas the data points for panels of AA2198 deviate significantly from this linear relation. In addition, there is a gap between the data points of flat and crenellated specimens, which is more pronounced at smaller values ofΔK.
The crenellated panel of AA2198 generally has lower crack propagation rates (the da/dN values) at the same ΔK compared to the flat panel. In the low ΔK range (< 15 MPa·m1/2), the da/dN values of AA2198 flat panel tend to be the same as those of the panels of AA2139. The same trend is also found in the high ΔK range (> 35 MPa·m1/2) where the da/dN values of all the panels tend to be the same.
During the fatigue tests, the crack opening load was measured along the crack path with a spacing of about 2 mm. Figure 4.5 a summarizes the averaged crack opening load measured in the flat and crenellated specimens of both materials, and the standard deviations of the data are shown in error bars. As it can be seen, both the flat and crenellated panels of AA2139 have the same averaged crack opening load of 46 kN. In comparison the flat specimen of AA2198 has a slightly increased crack opening load, whereas that of the AA2198 crenellated specimen is significantly higher. The scattering of the data is much larger in the AA2198 specimens compared to the AA2139 specimens.
The detailed variations of crack opening loads with half crack lengths of tested specimens are shown in Figure 4.5 b. Since the AA2139 flat and crenellated spec-imens have nearly the same trend of variation, only the data of the flat specimen is presented as reference for the clarity of the figure. As it can been seen, in the beginning (a < 30 mm) the crack opening loads of both flat specimens of AA2139 and AA2198 were at the same level, whereas in the same range that of the AA2198 crenellated specimen is significantly higher. When the half crack length exceeds 30 mm, the crack opening load of the AA2198 flat specimen increases dramatically up to the level of the AA2198 crenellated specimen. In the AA2198 crenellated specimen, there is a periodic variation of measured crack opening loads with the repeated repositioning ofδ5 clip gauge at the newest crack tips (P.1, P.2, P.3 . . . in
Figure 4.4. (a) The fatigue performance of flat and crenellated panels with differ-ent materials. (b)da/dN –ΔK plot of the four tested specimens.
4.1. Results 47
Figure 4.5. (a) Comparison of average crack opening load in the flat and crenel-lated specimens with AA2139 and AA2198 respectively. (b) The pro-files of crack opening load measured along the crack path.
Figure 4.5 b), which is schematically sketched in the lower part of the figure. When the clip gauge is just repositioned, the measured crack opening loads are among the highest values. Then with further extension of crack, the crack opening loads are observed to decrease significantly. Such a decrease of measured crack opening loads with increasing distance from the crack tip is also observed in the AA2198 flat specimen when the half crack length exceeds 40 mm. In the AA2139 flat specimen, the measured crack opening loads are relatively stable. No significant influence of the repositioning of the clip gauge is observed.
4.1.3 Fracture surface observation
Figure 4.6 compares the fracture surfaces of AA2139 and AA2198 panels at macroscopic scale. Both of the two AA2139 panels show flat crack surfaces with smooth single shear lips. The shear lips started to develop at positions marked by the white arrows and become stabilized in morphology at positions marked by the black arrows, where the whole section of the fracture surface is 45° inclined.
In the panels of AA2198, the fracture surface is much more tortuous due to the development of complex shear lip morphology. In the flat panel double shear lips firstly initiated at a half crack length of 30 mm, which are accompanied by a deviation of crack path around 10°. In contrast with the gradual rotation of the crack plane during the shear lip development in AA2139 specimens, abrupt changes of the shear lip plane are usually observed in AA2198 specimens as marked by the squares in Figure 4.6. In the crenellated panel, those abrupt changes are usually found associated with the thickness steps in the crenellations. In the crenellated panels very sharp shear lips are found along the crenellated side since the beginning of the fatigue tests. Those very sharp shear lips are less than 1 mm wide with the local crack surface about 70° inclined to the original crack plane.
At large crack lengths (a > 95 mm) stable single shear lips are observed in both flat and crenellated panels as those observed in the AA2139 panels.
The fracture surfaces of both materials can be generally divided into regions of tensile mode decohesion and regions of shear mode decohesion as shown in Figure 4.7. The shear mode decohesion is found always initiated at the specimen surface. In AA2139 specimens, after the initiation, the shear mode decohesion is found constrained in narrow stripes near both surfaces of the specimens for a considerable distance of crack extension. Meanwhile, with further extension of crack the whole crack surface becomes gradually inclined, which in the end has an angle about the 45° with respect to loading direction. During this period, the distinction between regions of tensile mode decohesion and regions of shear mode decohesion also becomes obscured. In contrast, in AA2198 specimens the shear mode decohesion, once it is initiated, will extend quickly across the whole crack surface except those very sharp shear lips formed along the crenellated side at small crack lengths. The borders between the tensile mode and shear mode decohesions are also much more distinct. In addition, in AA2198 specimens the inclination angles of the shear mode decohesion zones are usually found larger than in AA2139 specimens.
A closer examination of the crack surface in the tensile mode region under SEM shows a flat crack surface at the microscopic scale in both materials, which
4.1. Results 49
Figure 4.6. Fracture surface of (a) flat panels and (b) crenellated panels and the different morphology of shear lips developed at different positions.
Schematic sketches of the transverse slice of fracture surface are placed right above or below the positions where they are cut. The start point of shear lips and the position where they have been fully developed are marked by white and black arrows respectively. The abrupt changes of cracking plane are marked by rectangles.
Figure 4.7. The transitions from tensile mode decohesion to shear mode decohe-sion on the fracture surface of AA2139 and AA2198. The dashed lines indicate the local inclination of the fracture surface.
are compared with the corresponding microstructures revealed under the optical microscope (Figure 4.8). The irregularities on the surface are found mostly corre-lated with the positions of grain boundaries as indicated by the optical micrograph.
Within the grains the crack surface shows smooth appearance with visible stria-tions, the normal of which is found aligned well with the major loading direction.
Despite both materials show amorphous appearance in the shear mode region (the shear lips), there are two significant differences. Firstly, as shown in Fig-ure 4.9, abrasion zones near the sharp shear lips are usually observed in the AA2198 crenellated specimen, which are characterized by straight scratches in the direction perpendicular to the crack propagation direction. However, such abrasion zone is never seen in AA2139 specimens.
Secondly, in specimens made of AA2139, isolated islands of shear mode decohe-sion can be found in the region of tensile mode decohedecohe-sion (Figure 4.10). Those islands have higher density near the specimen surfaces and relatively lower density towards the center of the panels. With increasing crack length, the area fraction of those islands of shear mode decohesion increases. Then, instead of being small isolated regions they begin to form large bands, which finally dominate the whole fracture surface. This process accompanies with the gradual increase of inclination angle of the fracture surface at the macroscopic scale. When the inclination angle is approaching 45°, the fracture surface is dominated by shear mode decohesion with several islands of tensile mode decohesion distributed in between. In speci-mens of AA2198, such embedded distribution of shear mode decohesion are hardly seen, where the regions of different modes of decohesion are clearly separated with a distinct border.
4.1. Results 51
Figure 4.8. SEM micrographs of crack surfaces (b, c, e, f) of the two materials where the fatigue cracks grow in tensile mode and the comparison of crack surface morphology with the corresponding base material mi-crostructure (a, d).
Figure 4.9. SEM micrographs of the transition regions between the shear mode decohsion and the tensile mode decohesion on the crack surface of AA2198 (a) and AA2139 (b).
Figure 4.10. (a) Local shear mode decohesions (marked by arrows) occur in the tensile mode decohsion region on the crack surface of AA2139. (b) detailed view of the local shear-mode decohesions.